AI and Future Jobs
AI and Future Jobs
Estimates of Employment for 2030
Artificial intelligence (AI) dominates attention because it poses unprecedented threats of job displacement. Several recent forecasts have suggested that when AI matures, about 2025, almost half of present jobs will be lost, resulting in massive, permanent unemployment a few decades from now. Ray Kurzweil, now at Google, extrapolates the growth of computer power to estimate that a US$1000 PC will match the human brain about 2020. Ben Goertzel, the leader of the OpenCog project said "I am confident that we will have human-level AI by 2025. Maybe sooner."
This study addresses the looming issue of unemployment by forecasting the future distribution of jobs across the occupational spectrum. Following is a list of the major employment options using a conceptual framework of Manual, Service/Knowledge, and Creative work, as well as the options of a Guaranteed Minimum Income and Unemployment. We also present two alternative perspectives to stimulate thought, and we invite you to estimate these employment options as of the year 2030.
Thanks for participating. A tabulation of preliminary results is shown below. Curreent results can be seen after you complete the survey at the end.
The TechCast Team
The results as of March 5 are tabulated below, and suggest a remarkably reasonable path can be found through this difficult transition. Please download a full-size table at the end of this background to the survey.
A few conclusions stand out, suggesting an almost fully automated world increasingly valuing talents beyond knowledge.
Routine Jobs Lost About 21 Percent Fears of a 40- to 50 percent job loss seem exaggerated, at least in this time frame from 2015 to 2030. The main losses are in "Routine Work."
Job Gains in Creative Work The lost jobs are expected to be offset particially by an expected 10-percent gain in "Creative Work" -- entrepreneurs, leaders, artists, etc.
Unemployment Rise of Two Percent Unemployment is forecast to rise modestly by about 2 percent to reach 10 percent or so in 2030. This is serious but not a major crisis.
Guaranteed Minimum Income (GMI) Political presssures are likely to force governments to reduce social pain by introducing GMI benefits covering about 8 percent of the labor force. To be feasible, benefits must be means tested to exclude middle-class incomes.
The following analysis outlines the categories of work that are likely to occupy the labor force in OECD nations about 2030. Please note that this is a general framework that should apply to any economic system, and the categories distinguish between jobs that can and cannot be automated. For instance, automation also threatens undeveloped nations even more severely. But China, India, and the rest face a different situation at a lower level of development. In these nations, manual work is yielding to service/knowledge work as they develop a consumer economy. (Kurzweil AI, February 8, 2016)
Complex Manual Work Hard to Automate ~ 18% and stable
There always remains a significant portion of complex manual jobs that are hard to automateâ??barbers, janitors, farm workers, house cleaners, cooks, gardeners, repairmen, carpenters, dentists, care givers, etc. From 1982 to 2012, non-routine manual work in the US increased from 15 percent of employment to 19 percent. (MIT Review, Jan 8, 2016) Prof. Erik Brynjolfsson at MIT said these jobs "â?¦are not about to be replaced by machines [because they] involve sensory motor work, the skills of ideation, large-frame pattern recognition, and complex communication." (Tech Republic, Jan 8, 2015)
Manual work once dominated societies, but it has now been largely automated. This "residual" category of complex manual jobs employs about 18 percent of workers in OECD states now and likely to remain at roughly this level.
Routine Work Now Being Automated ~ 35% and Decreasing
Good AI and robotics is successfully automating this large group of routine, well-structured jobs that involve manual and service tasksâ??factory work, most clerical jobs, driving trucks, etc. Routine jobs fell from 56 percent of the US labor force in 1982 to 44 percent in 2012, and that proportion is certain to decrease over the next few decades as AI matures. (MIT Technology Review, Jan 8, 2016) TechCast estimates AI and robotics will replace 30 percent of routine knowledge work about 2025.
Service/Knowledge Work ~ 35% and Increasing
Teaching, law, medicine, management, the arts, and other professional work can be automated, but these jobs are likely to be transformed rather than eliminated or even reduced. They increasingly use AI to replace routine tasks and focus instead on the subjective, more complex tasks that machines can't do. Here are some facts to consider:
Cognitive Jobs Increasing This type of work grew from 29 percent in 1982 to 39 percent in 2012 in the US. (MIT Technology Review, Jan 8, 2016)
Most Jobs Only Revised A McKinsey study finds that fewer than 5 percent of occupations can be entirely automated using current technology. However, about 60 percent of occupations could have 30 percent or more of their constituent activities automated. (McKinsey Quarterly, Dec 2015)
Computerization Increases Jobs Studies show that occupations becoming more computerized can be eliminated, upgraded, and have new jobs created. Overall, automation actually grows occupations faster. (Atlantic, Jan 19, 2016)
Creative Work ~ 4% and Increasing
The new "creative economy" of leadership, entrepreneurship, innovation, collaboration, vision, etc., could grow dramatically because that is the next frontierâ??everything beyond knowledge. Some background trends:
- Hard to Automate A McKinsey study finds that capabilities involving creativity make up 4 percent of jobs in the US, but that is likely to grow. Creative work is central to the human experience and also difficult to automate. (McKinsey Quarterly, Dec 2015) Jobs that survive will involve abstract tasks in which workers have a comparative advantage of interpersonal interaction, adaptability, and problem-solving professional work, skilled labor (plumbers, builders, electricians, auto mechanics), customer-service relations, etc. (Business Insider, Aug. 25, 2013)
- New Industries The tech revolution is creating a flood of new products, services, and industries that are taking offâ??e-commerce, alternative energy, green economy, IoT, hi-tech homes, climate control, intelligent cars, etc. The field of energy, climate change, and environment alone is likely to create a US$10- to US$20-trillion global industry. TechCast estimates market saturation for about 50 technologies at an average of about US$1 trillion each, for a total of about US$50 trillion in new economic growth over the next few decades. That's as big as the present global economy. All these industries will create lots of new jobs, including routine jobs.
Guaranteed Minimum Income (GMI) ~ 0% and ?
Some propose to ease the unemployment problem by having governments guarantee everyone a minimum income. Added costs would be offset by savings from eliminating welfare, unemployment, and other benefits, and studies suggest no decline in working. A pilot in Canada, showed that a "Mincome" not only ended poverty but also reduced hospital visits and raised high-school completions. Experiments in Namibia cut poverty from 76 percent to 37 percent, raised education and health standards, and cut crime levels. One involving 6,000 people in India paid people US$7 monthâ??about a third of subsistence levels. It, too, proved successful. (Co.Exist, Mar 16, 2015)Switzerland is holding a referendum on a basic income of about US$2,800/month for all Swiss. (Independent, Jan 30, 2016)
Note that GMI benefits can be either "means-tested" or "universal." "Means-tested" limits benefits to those below a minimal income, while "universal" pays benefits to all citizens. In most nations, the political resolve to pay everybody may not be feasible. For instance, giving all adults in the US $10K/year ( < min wage) would cost about $3 trillion; this is as much as the entire Federal budget and would require doubling taxes. For this study, the GMI can be either means tested or universal, and it is considered an alternative to traditional unemployment with minimal benefits
Unemployed ~ 8% and ?
Without some form of GMI, the remainder of the labor force will be unemployed in the traditional sense of receiving only limited benefits. Global unemployment measures vary widely, ranging among OECD countries in 2014 from 3.6 percent in South Korea to 26.5 percent in Greece. The average was 7.3 percent, down from a peak of 8.3 percent in 2010 but still well above the 5.7 percent seen in 2007.
Worldwide, the most reasonable figures suggest a workforce of 3.36 billion people in 2014, of whom about 279 million were unemployed (8 percent.) At current growth rates, the labor force is expected to reach about 4 billion by 2025. At the rate seen in 2014, that woud leave 330 million unemployed.
- Forecast of 10% Unemployed by 2021 TechCast covers a Social Trend called Job Failure, which forecasts "Unemployment rises to 10 percent in OECD Nations" by 2021 +/- 3 years. The experts also estimate a social impact of -4, which is severe, and are 61-percent confident in their forecast.
- Prospects for Decline If automation displaces 40â??50 percent of workers, as many analysts believe it will, global unemployment could skyrocket to 1.6 billion, or about 20â??30 percent after job creation and other factors are taken into account.
- Prospects for Growth The logic of market economics suggests that AI will boost productivity, lower prices, and increase corporate sales and worker wages. This can be expected to increase demand throughout the economy, which would spur growth and create jobs. For instance, ITIF, a major think tank, is betting US$2,500 that unemployment in 10 years will be no higher than normal. (Christian Science Monitor, Dec 3, 2015)
Total Labor Force in OECD Nations = 100%
Note that percentages for the above categories have been adjusted to total 100 percent.
A Growth Perspective
New Jobs in a Higher-Order Frontier of Growth
William Halal, George Washington University and TechCast Global
Fears of mass unemployment by automation have been a constant throughout industrialization, but they are seldom realized. The evidence shows that automation reduces costs and frees up labor, which allows further economic growth and new jobs in areas of demand that were unexpected. Today's fears that AI will eliminate masses of jobs does not recognize how this dynamic will play out in the new economy that is emerging. They are somewhat reminiscent of the Y2K crisis that never materialized.
The key is to recognize that AI can automate knowledge work, but there exists a huge unexplored economic domain beyond knowledgeâ??creativity, entrepreneurship, vision, collaboration and diplomacy, marketing, supervision, and other higher-order functions that are uniquely human. See the figure, "Structure of Consciousness." Advanced AI may be able to solve tough problems, but it cannot provide vision, purpose, imagination, values, wisdom, and other capabilities that are essential for sound leadership and tough choices.
Yes, we can expect good virtual assistants to take over routine service tasks, but people will always want a real person to provide human contact. Staff is growing rapidly in universities, hospitals, research institutes, and other advanced settings for these reasons. The service and knowledge work sector could grow dramatically to 50 to 60 percent by 2030.
The "Singularity Myth" that advanced AI will surpass humans is as yet science fiction, and likely to remain so for our lifetimes. Jaron Lanier, a leading computer scientist at Microsoft, thinks the "AI Myth" that computers will be able to perform complex human tasks is unrealistic and leads to serious problems. (Edge, Nov 14, 2014)
There is scant evidence, if any, that state-of-the-art AI can duplicate advanced human abilities. Surveys regularly show about 90 percent or more think "There is a fundamental difference between machine intelligence and human intelligence." The idea of machines replacing higher-order abilities runs counter to experience of knowledgeable people.
The limits of machine capabilities may even grow more urgent. AI will mature, like all technologies, while society's demands are likely to escalate. The crises of IT system failures that damaged the US Government, Sony, Volkswagen, and hosts of other corporations could become move severe. IT systems even now frustrate people, and having advanced machines everywhere is going to pose far greater problems as the limits of AI press in. I can imagine a common experience of yelling at some "dumb machine."
The problem is that we have a hard time knowing what lies ahead in this new frontier. Who would have thought a few decades ago that most people today would do their work by staring into PC monitors, laptops, and mobile devices? There is no fixed amount of human endeavor, and work of different kinds will always appear to fill new economic demands. See the bullets in the Occupational Analysis above for examples of high-order work that will done in a wave of new industries.
People on the lower end of the normal IQ distribution may not be well-suited for hi-tech work. But they may do well among the 20 percent of the labor force that serve useful roles in the manual jobs that are hard to automate. Good employers will use knowlege-on-demand, online tutors, and other technologies to help those needing information. In principle, anyone can have special talents, and some low IQ employees could prove adept at service and creative work. With help and planning, they can be absorbed throughout the work force.
The value of machine intelligence pales in comparison to the utter complexity of human challenges looming ahead. A major frontier lies in addressing the massive global crises forming what TechCast has called a Global MegaCrisisâ??climate, energy, water, terrorism, financial instability, etc. These historic changes cannot possibly be handled by machines and will challenge humans to adapt for decades. See our webinar for more.
As always, this coming transformation will be an opportunity in disguise, as humanity is forced to create new institutions, learn to collaborate, and generally grow to an advanced state of global maturity. This will be a difficult transition, with global unemployment rising beyond 10 percent, possibly for years. Some nations will do better than others, depending on how well they address this challenge.
In the end, however, rather than diminishing people, the net effect of good AI may be to enhance the value of these higher-order talents that are a unique gift to humanity.
A Crisis Perspective
Employment Will Plummet as Human Jobs Are Automated
Jonathan Kolber and Owen Davies
Recent studies warn that by 2025 at least 40 percent of occupations could be automated out of existence. Bank of America Merrill Lynch, the University of Oxford, Brookings, and Nomura Research all have come to similar estimates. These findings rely only on technology now available or clearly on its way. No Singularity is required to cause mass job loss.
According to one estimate, half of major companies already are experimenting with AI. We believe this latest wave of automation strikes so deeply at human abilities that it will bring no compensating burst of job creation. The net effect is likely to be at least a 20-percent reduction in employment. Losses could be much greater.
Routine occupations are going away quickly. China's Foxconn has a fully automated factory that can operate 24/7 without ever turning on the lights. When autonomous vehicles are certified for travel, driving jobs will disappear; there are 3.8 million of them in the US alone. In Japan, companies are developing robots to replace RNs and home health workers. Guided by AI, there is little robots won't soon be able to do.
More ominously, automation is taking jobs that once required human versatility. Software manages huge investment portfolios. IBM's Watson diagnoses illness as well as top physicians and bests humans at the research employing 20 percent of lawyers. Narrative Science's computers write business and sports stories for major news outlets. AIs have found new mathematical proofs, developed patentable inventions, and even made at least one scientific discovery human researchers had missed. None of these activities qualifies as routine. Some require abilities that in a human being would be considered creativity.
AI becomes more capable by the day. Software recently has learned complex games simply by reading the rules, and the latest AI programs mimic human reasoning to beat flesh-and-blood competitors at chess and go. Microsoft's Azure software recognizes human emotions and responds appropriately, and Amelia responds to emotional cues in speech; Fortune 1000 companies use these tools to deliver better customer service. AI may not experience emotion, but it deals with it well enough to replace human workers in many functions.
We agree; there is a fundamental difference between human intelligence and the AI we know today. Yet, what matters are the results, not the means by which they are achieved. Decimating the job market does not require "general" AI that can replace every human function. It is enough that "narrow" AI can replace specific functions within limited domains of expertise. Watson is carefully marketed as an "assistant." Yet, the reality is that it will assist only those humans who remain after it has replaced the rest.
Optimists frequently cite Google as a hive of high-tech creativity and a promising example of things to come. Companies like Google need nowhere near as many workers as old-line manufacturers required. They cannot be counted on to replace the millions of jobs now at riskâ??certainly not in the few years available before unemployment goes viral. And most "creative" professions, like those at Google, require advanced mathematics and technical skills. Few of us have what it takes to prosper in such fields.
Some new jobs may appear, but they will not last for long. Machines have begun to learn by observation, by trial and error, and even from other machinesâ??as we do, but much faster. They are likely to master most new occupations before we humans ever have the chance. We face a time when humans will hop from one career to the next, struggling to stay ahead of automation. Saddled by debt and discouraged by a broken social contract, many may succumb to despair unless we find an alternative to endless retraining.
We are not entirely pessimistic about careers and employment. The best of us will still work as inventors, researchers, artists, and in whatever new occupations do emerge. Exceptional work will be rewarded in the professions. However, truly creative people within any field are always an elite. Most merely follow their lead, and machines soon will do it better.
Superstars will do fine in any system. The rest of us will need a new way to gain income and find meaning in our lives.
All Public Comments
© 2013-2017 TechCast Global Inc Printed By: Jun 01, 2020 For personal use only